Vol. 67 HYDROGENATING ACTION OF SHEEP-RUMEN CONTENTS 333Feuge, R. O., Cousins, E. R., Fore, S. P., Dupre, E. F. &

O'Connor, R. T. (1953). J. Amer. Oil Chem. Soc. 30, 454.Feuge, R. O., Pepper, M. B. jun., O'Connor, R. T. & Field,

E. T. (1951). J. Amer. Oil Chem. Soc. 28, 420.Garton, G. A. (1954). J. Sci. Fd Agric. 5, 251.Grass Driers' Association (1941). Analyst, 66, 334.Hartman, L., Shorland, F. B. & McDonald, I. R. C. (1954).

Nature, Lond., 174, 185.Hartman, L., Shorland, F. B. & McDonald, I. R. C. (1955).

Biochem. J. 61, 603.Hilditch, T. P. (1956). The Chemical Constitution of Natural

Fats, 3rd ed. London: Chapman and Hall, Ltd.Hilditch, T. P., Patel, C. B. & Riley, J. P. (1951). Analyst,

76, 81.Hoflund, S., Holmberg, J. & Sellman, G. (1956). Cornell Vet.

40, 53.James, A. T. & Martin, A. J. P. (1952). Biochem. J. 50,

679.Klenk, E. & Bongard, W. (1952). Hoppe-Seyl. Z. 290,

181.Lemon, H. W. (1949). Canad. J. Res. B, 27, 605.Markley, K. S. (1947). Fatty Acids, 1st ed. New York &London: Interscience Publishers.

Marvel, C. S. & Rands, R. D. jun. (1950). J. Amer. chem. Soc.72, 2642.

Rebello, D. & Daubert, B. F. (1951). J. Amer. Oil Chem. Soc.28, 177.

Reinbold, C. L. & Dutton, H. J. (1948). J. Amer. Oil Chem.Soc. 25, 120.

Reiser, R. (1951). Fed. Proc. 10, 236.Reiser, R. & Reddy, H. G. R. (1956). J. Amer. Oil Chem.

Soc. 33, 155.Shorland, F. B. (1952). J. appl. Chem. 2, 438.Shorland, F. B. (1955). Progress in the Chemistry of Fats and

Other Lipids, vol. 3, p. 275. London and New York:Pergamon Press.

Shorland, F. B. (1956). Food Manuf. 31, 272.Shorland, F. B. & De La Mare, P. B. D. (1945). J. agric. Sci.

35, 33.Shorland, F. B., Weenink, R. 0. & Johns, A. T. (1955).

Nature, Lond., 175, 1129.Society of Public Analysts (1933). Report of the Sub-Committee on the Determination of UnsaponifiableMatter in Oils and Fats, Analyst, 58, 203.

Swern, D. & Parker, W. E. (1953). J. Amer. Oil Chem. Soc.30, 5.

Warner, A. C. I. (1956). J. gen. Microbiol. 14, 733.Weenink, R. 0. (1956). Nature, Lond., 178, 646.Willey, N. B., Riggs, J. K., Colby, R. W., Butler, 0. D. &

Reiser, R. (1952). J. Anim. Sci. 11, 705.

The Ninhydrin Reaction and its Analytical Applications

BY H. MEYERDepartment of Biological and Colloid Chemi8try, The Hebrew Univer8ity, Jerusalem, Israel

(Received 25 January 1957)

The colorimetric or spectrophotometric measure-ment of the absorption produced by the purplereaction product (ANP) between ac-amino acids andninhydrin at 570 m, has become one of the mostwidely used methods for the quantitative determi-nation of small quantities of a-amino acids. Itsanalytical application is, however, sometimesadversely affected by one or more of the followingfactors. Since theANP has not been prepared in thesolid state, its extinction coefficients could not bedetermined; in consequence it has been necessary toexpress analytical results in terms of 'colour yield'in relation to that given by known quantities of areference standard reacting under defined con-ditions (Moore & Stein, 1948). It has often beenobserved that the ANP yield was not constantowing to the occurrence of oxidative side reactionsdecomposing it to colourless products. Moore &Stein (1948) sought to overcome this effect by theaddition of reducing agents like stannous chlorideor hydrindantin to the reaction systems. The ab-sorption maximum of the ANP was found to varywithin the 500-580 mu range, thus making im-possible absorption measurements at a fixed wave-

length. A method devised to eliminate this effect,by treating with alkali spots ofANP on chromato-grams, was described by Kay, Harris & Entemnan(1956).The sensitivity of the reaction, i.e. the colour

yield by a given amino compound produced underdefined conditions, is significantly influenced bythe composition of the solvent in which the reactiontakes place. It is highest in organic solvents anddecreases in solvent mixtures containing water(Yanari, 1956).The experiments reported in this paper were

designed to investigate the properties and be-haviour ofANP, especially its absorption spectrumand its changes under different conditions and tocorrelate the changes in ninhydrin reactivity withmeasurable properties of this compound and, inturn, with structural changes undergone by theninhydrin molecule in different solvents.

MATERIALS AND METHODSAll chemicals used were of analytical grade. The disodiumsalt of ethylenediaminetetra-acetic acid (EDTA) was re-crystallized from water before use. Hydrindantin was

334 H. MIprepared according to West & Rinehart (1942) by reductionof ninhydrin with ascorbic acid.

Solvents. The absorption spectra of ninhydrin and hydrin-dantin were measured in absolute ethanol. In all otherexperiments 95-5% (w/v) ethanol was used. All water usedwas twice distilled. Spectrophotometric measurementswere performed in a Beckman quartz spectrophotometermodel DU, in 1 cm. cells.

Preparation of ANP. A solution of 0-5 g. of DL-alanine,0-2 g. of ninhydrin hydrate and 0-1 g. of EDTA in 50 ml. ofwater was refluxed for 8 min. and cooled to room temp.Small quantities of bis-diketoindanyl (Moubasher &Ibrahim, 1949) were filtered off. The filtrate was then cooledto 2° and acidified with cold (20) 20% (w/v) H2SO4. Thereddish precipitate formed was filtered off and washed withice-cold water as rapidly as possible, until the filtrate showeda strong purple colour. The precipitate was then sucked dryand suitable fractions of the still slightly wet material weretaken for nitrogen determination and spectrophotometricmeasurements. Another weighed portion was dried toconstant weight at 105°. All analytical results were calcu-lated with this dry-weight determination. The nitrogencontents of a wet sample, recalculated to dry weight, andthat determined after previous drying were identical, sothat no loss of material had occurred on drying (Found:N, 4-5; 4-7. C18H.04N requires 4.6%). It was, however,found that a loss of 10-12% of absorption occurred ondrying, indicating that a colourless non-volatile substancehad begun to form, presumably hydrindantin (Moubasher &Ibrahim, 1949). The absorption spectrum is shown in Fig. 1.Ninhydrin blue. A solution of the blue compound formed

by ninhydrin with concentrated alkali (ninhydrin blue)suitable for spectrophotometric measurements was pre-pared by dissolving 13 mg. of ninhydrin hydrate in 3 ml. of6N-NaOH. A blue colour is rapidly formed which is stableto air and heating, but disappears when the solution isdiluted with water.

Hydrindantin blue. A solution of the blue compoundformed by hydrindantin with alkali (hydrindantin blue) wasprepared by dissolving 10 mg. of hydrindantin in 25 ml. of0- N-NaOH. For spectrophotometric measurements thesolution was rapidly transferred to the absorption cell andcovered with a layer of liquid paraffin to protect it from air.

General procedure. The reaction mixture, consisting of0-1 ml. of an aqueous solution of the nitrogen compound tobe tested, 0-2 ml. of 0-3% aq. EDTA soln. and 4 ml. of0-2% ninhydrin soln. in the solvent employed in therespective experiment was refluxed for 20 min. in a25 mm. x 200 mm. Pyrex test tube fitted with a cold-fingercondenser reaching to within 3 cm. above the boilingsolution. The cooled solution was filtered, if necessary, fromcrystals of EDTA through a small wad of cotton wool andmeasured in the spectrophotometer at 578 or 408 mu, ifnecessary after dilution with an appropriate volume ofsolvent. The reagent blank produced by ninhydrin withrecrystallized EDTA is negligible when measurements aretaken at 578 m., but at 408 mp the reagent blank issignificant owing to the absorption of ninhydrin at thiswavelength (see Fig. 2).For experiments in which the dependence of ninhydrin

reactivity with amino acids upon the pH of the reactionmixture was determined: reaction mixtures of constantninhydrin content and varying pH were prepared bybringing 25 ml. of a 0-4% aq. ninhydrin soln. to approxi-

AYER I957mately the desired pH by the addition of 0-1 -HCl orNaOH and adding water to make the total volume 50 ml.The final pH of 4 ml. of this solution, to which 0-1 ml. ofwater containing 500pg. of DL-alanine but no EDTA wasadded, was then accurately determined.For experiments to test the availability of the 'general

procedure' for the quantitative determination of aminonitrogen 95-5% (w/v) ethanol was used as the solvent.

(a) Determinations of colour yield at 578 and 408 m/L atdifferent concentrations were carried out for all the naturalamino acids and glycylglycine, glycyl-leucine, leucylglycine,triglycine and leucylglycylglycine. They were added asaqueous solutions containing from 0-2 to 9,ug. of mc-aminonitrogen/0-1 ml.

(b) The influence of buffers normally used was determinedin a parallel series ofexperiments, under the same conditionsexcept that the amino acids and peptides were dissolved inacetate, or citrate or phosphate buffers (0- 1M, pH 6-5).

(c) The reactivity of nitrogen-containing buffers wastested by adding 5 mg. of 2-amino-2-hydroxymethylpro-pane-1:3-diol (tris) or sodium diethylbarbiturate in 0-1 ml.ofwater to a reaction mixture not containing other nitrogencompounds.

(d) Ninhydrin reactivity with free ammonia, methyl-amine, ethylamine and ethylenediamine was tested inreaction mixtures containing quantities of each nitrogencompound equivalent to 10 jg. of nitrogen. Ammoniumchloride and the hydrochlorides of the amines mentionedwere added in quantities equivalent to 2 mg. of nitrogen.

All experiments mentioned under (a) and (d) were alsocarried out with hydrindantin instead of ninhydrin.

Kinetic experiments. The desired amount of DL-alaninewas dissolved in 4 ml. of water, 8 ml. of a 0-3% aq. EDTAsoln. was added, followed by 160 ml. of ethanol containingthe desired amount of ninhydrin. This reaction mixture wasbrought to the boil in a cylindrical separating funnel of250 ml. capacity, externally wound with electrical-heatingtape and fittedwith a reflux condenser. Thereaction mixturewas stirred by a vigorous stream of nitrogen gas. Samples(3 ml.) were withdrawn through the tap of the funnel intotest tubes kept in ice-water at 20 sec. intervals during thefirst 2 min. of reaction time (counting from the beginning ofboiling), at 30 sec. intervals during the following 6 min., andthen every 3 min. until the end of the reaction time (50 min.).Spectrophotometric measurements were taken at 578 and408 mu; if necessary, the samples were diluted with anequal volume of ethanol. All reactant concentrations aregiven in Table 1.

RESULTSPropertie8 of ANP. In the solid state ANP ia

a reddish substance, easily soluble in all solventscontaining alcoholic or phenolic hydroxyl groupsand insoluble in ether, chloroform and non-polarorganic solvents. Attempts to recrystallize it fromwater failed owing to decomposition on warming.It completely decomposes in dilute aqueous solutionif kept for some hours, and almost immediatelywhen such solutions are acidified and slightlywarmed. Alkaline ANP solutions and those con-taining 0- 1 % of EDTA are stable for some days.Solutions of purified ANP in ethanol or n-butanol,stabilized by the addition ofEDTA, can be kept for

E

NINHYDRIN REACTIONseveral weeks. A characteristic property of ANP,which can also be used to purify the compound, is itsspecific absorption by anion exchangers. The effectcan easily be demonstrated by shaking aqueousANP solutions with De-Acidite or AmberliteIR-4B in their OH- forms. Elution can be effectedwith aq. N-NaCl soln. or with ethanol. The nitrogencontent of the ANP fits the structures proposed byRuheman (1911) (I) and by Woker & Antener(1937) (II). Both (I) and (II) are assumed to beanionic structures (Clarke, 1948; Wieland, 1949).

COl \COOH

(I)

0/\a a&DcCO CO

(II)

The absorption spectrum of pure ANP shows twowell-defined bands, with Ama 578 and 408 mu andE1oy. 660 and 724 respectively (Fig. 1, curve 2). Themaximum at 408 m/h is especially characteristic forANP, since it does not appear in the absorptionspectra of the nitrogen-free compounds ninhydrinblue and hydrindantin blue (Fig. 5) which bothabsorb strongly in the 570-580 m,u region. Theabsorption maximum at 408 mu does not form apart of the absorption spectrum ofmurexide, whichis sometimes regarded (Clarke, 1948) as a structuralanalogue ofANP (Fig. 1, curve 3).

Since the extinction coefficients of ANP havebeen determined, it can be decided whether ANP

3-0

2-5

2-01

2

%~~~F,41%.~~~~ '

-' ~~~~ j1

/ ' \T\ .7,~~~~~~1J

600 500 400 300Wavelength (mp)

Fig. 1. Absorption spectra: 1, ANP (10 mg./l.) inn-butanol;2, ANP (10 mg./l.) in ethanol or water, containing 1%(v/v) of 0-3% aq. EDTA soln.; 3, murexide (10 mg./l.) inwater.

contains all the nitrogen originally present in theamino compound from which it had been formed.A solution of 9-2 mg. of ANP/l. gave E:L at578 m,u 0-600. The same extinction was measuredin an ANP solution in which the ANP had beenformed by the reaction of 1 1-62 jig. of alanine withninhydrin. The reaction mixture (total vol. 4-3 ml.)contained therefore 1-83 ,ug. of amino nitrogen,which corresponds, calculated to a nitrogen contentof the ANP of 4-62 %, to an ANP concentration of0-2 mg./l. All the alanine nitrogen has thereforebeen converted into ANP nitrogen.The absorption spectrum of ANP formed in a

reaction mixture in n-butanol in the completeabsence of water is shown in Fig. 1, curve 1. It canbe seen that absorption at both characteristicwavelengths is depressed, but enhanced at allothers up to 530 miu. The colour of the materialproduced in butanol is reddish instead of the deep-purplish blue. When minute amounts of water arepresent in the reaction mixture, absorption spectraare produced which vary in appearance between theextreme shown and that of the normal ANPspectrum. When the water content of the reactionmixture reaches or exceeds 5 %, the normal ANPspectrum is always produced. Since the existence ofan anionic structure in the ninhydrin ANP moleculeis probable, it seems reasonable to assume that thenormal ANP spectrum is due to the existence of theANP anion, whereas the spectra produced in theabsence of water are characteristic of the undis-sociated form ofANP.

Kinetic8 of the formation of ANP. The results ofthe kinetic experiments are given in Table 1. Colourformation was preceded by a latent period duringwhich no colour was formed. The colour-formationreaction itself, when measured at any of the nin-hydrin concentrations given in Table 1 and atalanine concentrations varying from 0-0129 to0-103 m-mole/l., followed (within an experimentalerror of 5 %) unimolecular kinetics, the rate ofANP formation being independent of the initialalanine concentration. Measurements taken atconstant alanine and varying ninhydrin concentra-tions showed, however, that both the latent periodsand the reaction half-times were not independent of

Table 1. Kinetic data on the formation ofANPby the reaction of ninhydrin with DL-alantne

Alanine concentration was 0-0258 m-mole/l.Concn. of Molar concn. Latent Reactionninhydrin ratio period half-time(m-moles/l.) (alanine: ninhydrin) (sec.) (min.)

5-82 1:225 61 511-64 1:450 30 2-517-76 1:675 15 2-022-38 1:900 6 1-756-56 1:1800 6 1-7

Vol. 67 335

H. MEYERninhydrin concentration, as should be expectedwith the large excesses of ninhydrin used (seeTable 1). It can be seen that both latent period andreaction half-time become independent ofninhydrinconcentration only when the molar-concentrationratio alanine:ninhydrin equals or exceeds 1:900.This result seems to show that, of the total nin-hydrin concentration present, only a small part isavailable for colour formation.

1-0

0-5bO

0

- --A 3!5/ /<

/ 1/ 1

4/ 1,2/ 3/

lI / l / l

400Wavelength (mp)

500

Fig. 2. Absorption spectra: 1, ninhydrin (1 g./l.) inabsolute ethanol; 2, hydrindantin (1 g./l.) in absoluteethanol; 3, ninhydrin (1 g./l.) in water; 4, ninhydrin

,(1 g./l.) in aq. NaOH soln. (pH 10.25).

0-6

0-5

0-4 -

0:0-3 -

0-2 -

0.1 o\

w 50 100Water content of solvent (%)

Fig. 3. Changes of absorption of ninhydrin at 356 mp (1)and of the colour value of the product of its reaction withamino acids (2) with increasing water content of thesolvent. For (2), 10,tg. of DL-alanine was used in 0'2%ninhydrin soln.; other conditions are as described underMaterials and Methods.

Factor8 affecting the reactivity of ninhydrin.Success in applying the ninhydrin reaction toanalytical problems depends to a large measureupon the state of this compound in solution, sincethe sensitivity of the reaction is much greater inorganic solvents than in water. Another factorgreatly affecting sensitivity is the HI ion concentra-tion (McFadyen, 1948), especially in aqueoussolution. As ninhydrin solutions are known todeteriorate with time, a method of standardizingsuch solutions to constant activity would bedesirable.The following experiments were an attempt to

provide evidence for the existence of an activeform of ninhydrin, the concentration of whichvaries with different conditions and determines thesensitivity of the ninhydrin reaction. It was foundthat reactivity changes of ninhydrin, especiallythose occurring upon changing the composition ofthe solvent, can be correlated with changes in itsabsorption spectrum happening at the same time.The absorption spectrum of ninhydrin in ethanol isshown in Fig. 2, curve 1. The absorption maximumat 356 m,u almost completely disappears when thesolvent is water (Fig. 2, curve 3). When absorptionat 356 mp was measured in ninhydrin solutionswhere the ethanol was gradually replaced by water,it could be shown that the decline in absorption islinearly proportional to the increase in water con-tent (Fig. 3, curve 1). Concurrently, the colouryield obtained with alanine also declined in linearproportion with increasing water content of thesolvent (Fig. 3, curve 2).

Since both absorption at 356 miz and the ANP-producing activity of ninhydrin are linearly pro-portional to the water content of the solvent, alinear-functional relationship exists between thetwo factors. The colour-producing activity ofninhydrin in any solvent mixture can therefore beexpressed in terms of the extinction coefficientmeasured at 356 m,. Ninhydrin solutions can thusbe standardized to their maximum or any givenpercentage activity, either by the addition of waterto the ethanolic ninhydrin solutions or by varyingthe ninhydrin concentration.The 0-2 % ninhydrin soin. in ethanol generally

used in the experiments reported in this paper wasstandardized to an extinction value of 1-00 at356 m,u, measured in a 1 cm. cell.Apart from the composition of the solvent, the

sensitivity of the ninhydrin reaction is most in-fluenced by the H+ ion concentration in the solvent.In aqueous solutions with a pH ofbelow 4 no colourformation took place. Reactivity, as expressed bycolour yield, increases sharply to reach its peak atpH 6-18, declined then to half its maximum valueat pH 8-8 and disappeared completely at pH 10 andabove (Fig. 4, curve 1). The absence of reactivity in

336 I957

LLJ

NINHYDRIN REACTION

acid solutions is due to the instability of ANP inacid. When solutions containingANP were acidified,theANP colour disappeared almost instantaneouslyin solutions ofpH of below 4 and more slowly in lessacid solutions. It can therefore be concluded that insuch solutions the colour yield observed depends on

the ratio between the rates of formation and de-struction of ANP. No such effect is possible inalkaline solutions in which ANP is stable; reactivitychanges in alkaline solutions have therefore to beexplained by other effects, such as changes in theninhydrin molecule. Evidence for such an effect canbe seen in the significant changes occurring in theninhydrin-absorption spectrum in alkaline solution,in contrast with acid conditions where no changes inthe absorption spectrum of ninhydrin could bedetected at different degrees ofacidity. A ninhydrinsolution in water to which increasing quantities ofalkali are added turns yellow and shows a new

absorption maximum at 380 m,u. Extinction at thiswavelength reached half its maximum value atpH 8-8 and its maximum value at pH 10-25 (Fig. 4,curve 2). (The shape of the absorption spectrum atthis point is shown in Fig. 2, curve 4.) Extinction at380 mp, declines then rapidly and disappearscompletely at pH 12 and higher.The potentiometric titration curve of ninhydrin

(Fig. 5) shows that ninhydrin behaves like an acidwith a pK of 8-8. When the alkali concentration ofaqueous solutions containing ninhydrin is increasedto 6N, the colour of ninhydrin blue appears. Itsabsorption spectrum (Fig. 6) shows, in addition tothe maximum at 580 m,, a pronounced absorption

maximum at 356 m, resembling that shown byninhydrin in ethanol (Fig. 2, curve 1). It should benoted that the absorption spectrum of ninhydrinblue is very similar to that of hydrindantin blue,a compound formed when hydrindantin, the re-

duction product of ninhydrin, is dissolved in01N-NaOH (Fig. 6).Hydrindantin has been shown to react with

amino acids in an aqueous medium and to yield thesame reaction products, including ANP (Moubasher& Ibrahim, 1949). We found that in ethanolicsolution hydrindantin produces amounts of ANPequal to those obtained with ninhydrin undersimilar conditions.

Quantitative determination of amino acid8

Complete conversion of the a-amino nitrogen intoANP nitrogen, i.e. a theoretical colour yield, was

obtained with all the amino acids and peptides

9'

E 6

la 7

v 53104m 3I

2Z I l I I I I

S 7 8 9 10 11 12pH

Fig. 5. Titration curve ofninhydrin(4-92 m-molestitrated).

v

CI

L.1v

0L

pHFig. 4. Changes in the colour-producing activity of nin-

hydrin with amino acids (1) and of its absorption at380 mu (2) with pH in aqueous solution, expressed as

percentages of the respective maximum values. Ab-sorption of ninhydrin was measured in 0-2% soln. andcolour values were determined with 500 ,ug. of DL-

alanine in 0-2% aq. ninhydrin soln.; other conditions as

described under Materials and Methods.

22

W-NA10

/ | | | | |I~~~"

600 500 400 300Wavelength (mpZ)

Fig. 6. Absorption spectra: 1, hydrindantin blue (fromlOmg. of hydrindantin in 25 ml. of 0O1:-NaOH); 2,ninhydrin blue (from 13 mg. of ninhydrin in 3 ml. of6N-NaOH); 3, hydrindantin in ethanol (1 g./l., freshlyprepared solution).

Bioch. 1957, 67

Vol. 67 337

tested, with the exception of proline, hydroxy-proline, phenylalanine, tyrosine and tryptophan.Hydroxyproline and proline cannot be determinedby any method measuring ANP production, sincetheir reaction product with ninhydrin differs instructure and absorption characteristics from ANP(Grassmann & Arnim, 1934). Phenylalanine,tyrosine and tryptophan yield about 70% of thetheoretically expected colour value. They can,however, be determined with the use of specialcalibration curves, since the colour values given bythem were reproducible. In all other cases theconstruction of a single calibration curve is suffi-cient, or the a-amino nitrogen content of the sampleanalysed can be calculated directly by

pg. of a-amino nitrogen in sample= 10OAva/(E 1 %. x 1),

where A is the extinction (optical density) valuemeasured for the sample, v is the total volume of thereaction mixture (4 3 ml.), a is the nitrogen contentof ANP (in %), ElC is the specific extinctioncoefficient of ANP at the wavelength used and 1 isthe cell length in cm.

It should be noted that the method is strictlyspecific for a-amino nitrogen; lysine, arginine,ornithine, citrulline and histidine were all found toreact with their oc-amino groups only.

Strictly linear dependence of extinction (opticaldensity) on concentration was observed withsamples containing amounts of oc-amino nitrogenranging from 0 5 to 6 pg. The accuracy ofthe methodis ± 3 %. The same results were obtained when thecompounds tested were dissolved in acetate,citrate or phosphate buffers. The nitrogen-contain-ing buffers tris and diethylbarbituric acid did notreact with ninhydrin.Both ammonia and the amines tested, when

present as free bases, reacted in ethanolic solutionwith ninhydrin. They yielded 17-30% of theamount of ANP theoretically expected, but theresults obtained varied within very wide limits. Theinterference of ammonia and amines with thedetermination of amino acids can, however, easilybe eliminated by converting them into their salts,since ammonium and amine salts did not yieldmeasurable amounts of colour.

DISCUSSION

Structure of ninhydrin. Since it has been shown(Fig. 2) that the absorption spectra ofninhydrin andhydrindantin are the same in the region most con-nected with activity changes, it may be concludedthat both compounds owe their reactivity withamino compoumds to similar structures. Accordingto its discoverer, Ruheman (1911), hydrindantin isa condensation product of ninhydrin (V) with its

I957reduction product 1:3-dioxo-2-hydroxyindane andhas the structure (III). Schonberg & Moubasher(1949) favoured structure (IV). Hydrindantin does

CO H OH CO(III)

(IV)

not therefore contain the carbonyl group in the2-position to which the reactivity of ninhydrin (V)with amino compounds is generally believed to bedue (Clarke, 1948). It is therefore difficult to see howhydrindantin could react with amino compounds atall, and why ninhydrin, usually represented as (V),should show exactly the same absorption character-istics in the 300-400 m,u region as its reductionproduct, hydrindantin.

co ~ --~Co2CO a C

(V) (V a)

The system (V a), common to both hydrindantinand ninhydrin, cannot be responsible for this effectsince the absorption spectrum of ANP, which alsocontains (V a), does not show any absorptionmaximum at 356 m,i. It is therefore assumed thathydrindantin, on dissolving in organic solvents,loses one molecule of water thereby forming struc-ture (VI), and that ninhydrin forms a similar

CCo \/ C/ D

(VI)

/C\II\(VII)

structure (VII) by losing two molecules of water.This structural change is in hydrindantin accom-panied by absorption changes. Freshly preparedhydrindantin solutions show a red colour with an

338 H. MEYER

NINHYDRIN REACTIONabsorption maximum at 510 mu (Fig. 6, curve 3).When kept the colour fades and the absorptionspectrum then assumes the form shown in Fig. 2,curve 2. Absorption at 356 mH is due to the presenceof (VI) and (VII), which are the reactive forms ofhydrindantin and ninhydrin respectively, butappear in solution only in small equilibrium concen-

trations. This assumption is borne out by the factthat a very large excess of hydrindantin and ofninhydrin is necessary to make them react withamino compounds. The dependence of bothabsorption at 356 ml, and reactivity of ninhydrinbecome now easily explainable by assuming that, inthe presence of water, (VII) is split into two mole-cules of (non-reactive) ninhydrin hydrate. Evidencefor the existence of a condensed form of ninhydrincan be seen in the fact that hydrindantin has beenprepared by Schonberg & Moubasher (1949) by theaction of ninhydrin and magnesium in the presence

ofmagnesium iodide in ether. This is essentially themethod for the synthesis of pinacols from aromaticketones published by Gomberg & Bachmann (1927),which is believed (Wheland, 1953) to lead throughthe intermediate stage of the often highly colouredmetal ketyls. It is suggested that the deep-bluecompounds ninhydrin blue and hydrindantin blueboth formed by the action of alkali on the parentcompounds could be regarded as analogous to themetal ketyls. The fact that ninhydrin blue has an

absorption spectrum very similar to that ofhydrindantin blue (Fig. 6) can also be taken as

evidence for the existence of a condensed form ofninhydrin, since hydrindantin blue is most probablyto be regarded as a derivative of the condensedstructure (IV).

Simple reaction mechanisms for explaining theninhydrin reaction, the reactivity of amino com-

pounds with ninhydrin and hydrindantin, can beconstructed with the use of structures (VI) and(VII), the complete oxidative deamination of theamino compounds being performed by one moleculeof ninhydrin or two molecules of hydrindantin,leading to the common intermediate (VIII) fromwhich either (I) or (II) could be formed.

CO NH2 OH CO

(VIII)

Structure of ninhydrin in aqueous solution. Thegreatly diminished reactivity of ninhydrin inaqueous solution, due to the prevalence of thenon-reactive form (V), is still further reduced andfinally abolished when such solutions are madealkaline (Fig. 4). This effect can be explained by the

assumption that ninhydrin (V), which in aqueoussolution can still form a small amount of the activeform (VII) of ninhydrin, can not do so if convertedby alkali into the univalent ion (IX). This ion is

co0- ~co 0-a/\ aC c,

co OH 00 0-(IX) (X)

capable of forming resonance structures and there-fore shows absorption (at 380 m,). At pH 8-8, thepK of the acid form of ninhydrin, ninhydrin re-activity has half the value it has under conditionsexcluding the presence of (IX). Concurrently,absorption at 380 m,, which is due to the formationof (IX), has half its maximal value at the same pH.At pH 10-25 all ninhydrin has been converted into(IX); accordingly, absorption at 380 m, is maxi-mum and reactivity is completely abolished. Whenthe solution is made still more alkaline, the bivalention (X) is formed which is incapable of formingresonance structures. There is then no absorption at380 m,u (Fig. 4).

Analytical conclusions

The high sensitivity of the ninhydrin reaction canbe fully utilized when ethanol is used as the solvent,since in ethanol the formation of the active form ofninhydrin (VII) is favoured and the formation ofits inactive forms (V), (IX) and (X) is largely sup-pressed. Sensitivity is not measurably diminishedby the amount of water present, which is necessaryto avoid the formation of atypical ANP spectra andto hold a sufficient amount of EDTA in solution.Ethanol also largely obviates any interferencelikely to be caused by the pH of the solution to beanalysed, except in extreme cases. Solutions con-taining the amino acid to be determined may varyfrom pH 4-5 to 8. When ammonia or amines arepresent, care should be taken to neutralize themcompletely.

Interfering factors. Strong oxidizing or reducingsubstances destroy the ANP. Trichloroacetic acidseriously interferes with ANP formation and musttherefore quantitatively be removed by etherextraction from solution to be analysed. The inter-ference caused by metal ions present in the concen-trations usually found in biological systems iseliminated by the use of EDTA. No other inter-fering factors have so far been found.The method described is primarily intended for

the determination of amino acids and peptides incolumn eluates or cell extracts. It can be adaptedto the determination of amino acids on paperchromatograms; the zones containing the amino

22-2

Vol. 67 339

340 H. MEYER I957acids are cut out and refluxed in a reaction mixtureconsisting of 4 ml. of ninhydrin solution and 0-2 ml.ofEDTA solution and 0- 1 ml. of water.The method has been so far successfully employed

for the determination ofamino acids in ether-treatedtrichloroacetic acid and warm-water extracts ofyeast cells (Benziman, 1956) and for following theenzymic decomposition of polylysin (M. Rigbi,personal communication).

SUMMARY

1. The coloured reaction product of amino acidswith ninhydrin (ANP) was prepared in the solidstate. Extinction coefficients for its absorptionmaxima at 578 and 408 m,u were determined.

2. By using the fact thatANP is greatly stabilizedin the presence of ethylenediaminetetra-acetic acid,uniform absorption spectra are obtained and oxi-dative side reactions leading to the decomposition ofANP are avoided.

3. The a-amino nitrogen of amino acids isquantitatively transferred to ANP.

4. The changes in the reactivity of ninhydrinwith amino compounds in different ethanol-watermixtures and in aqueous solutions at differentH+ ion concentrations, and the accompanyingchanges in its absorption spectrum, are explainedby structural changes in the ninhydrin molecule.

A structure for the reactive form of ninhydrin andhydrindantin is suggested,

5. An analytical method for the spectrophoto-metric determination of 05-6 jLg. of cc-aminonitrogen, contained in amino acids or peptides, isdescribed.The author wishes to express his gratitude to Professor

A. Neuberger, F.R.S., for helpful advice and discussions.

REFERENCES

Benziman, M. (1956). Ph.D. Thesis: The Hebrew Uni-versity, Jerusalem.

Clarke, H. T. (1948). Organic Chemi.try, 2nd ed., vol. 2,p. 1049. Ed. by Gilman, H. New York: Wiley.

Gomberg, M. & Bachmann, W. E. (1927). J. Amer. chem.Soc. 49, 236.

Grassmann,W. &Arnim,K. v. (1934). LiebigsAnn. 509,288.Kay, R. E., Harris, D. H. & Entenman, C. (1956). Arch.

Biochem. Biophy8. 63, 14.McFadyen, D. A. (1948). J. biol. Chem. 153, 507.Moore, S. & Stein, W. H. (1948). J. biol. Chem. 176, 367.Moubasher, R. & Ibrahim, J. (1949). J. chem. Soc. p. 702.Ruheman, J. (1911). J. chem. Soc. 99, 797.Schonberg, A. & Moubasher, R. (1949). J. chem. Soc. p. 212.West, E. S. & Rinehart, R. E. (1942). J. biol. Chem. 146,105.Wheland, G. W. (1953). Advances in Organic Chemistry,2nd ed., p. 717. New York: Wiley.

Wieland, T. (1949). Fortschr. chem. Forsch. 1, 211.Woker, G. & Antener, I. (1937). Helv. chim. acta, 20, 1260.Yanari, S. (1956). J. biol. Chem. 220, 683.

Blood Lipids1. PLASMA LIPIDS OF THE LACTATING COW: CHROMATOGRAPHY ON SILICIC ACID

BY G. A. GARTON AND W. R. H. DUNCANRowett Research In8titute, Buck8burn, Aberdeen8hire

(Received 13 March 1957)

Arteriovenous studies (for review see Folley, 1956)have demonstrated that no component of bloodlipids except triglycerides and possibly cholesterolesters is absorbed from the blood plasma by thelactating ruminant udder. It is therefore of someimportance to study the fatty acid composition ofthese classes of lipid, though it is first necessary toseparate them from each other and from otherplasma lipids.The separation of mixtures of lipids into their

component classes was restricted for many years totheir segregation into acetone-insoluble lipids(phospholipids) and acetone-soluble lipids (gly-cerides, sterol esters, free sterols and fatty acids),from which mixture the last two components could

each be recovered chemically. However, no meansof distinguishing between fatty acids in combina-tion as glycerides and as sterol esters was deviseduntil Kelsey (1939) used pig pancreatic lipase forthe selective hydrolysis of glycerides in the presenceof cholesterol esters. This procedure was employedby Kelsey & Longenecker (1941) in an investigationof the component fatty acids of the plasma lipids oflactating cows. As has been discussed by Lovern(1956), it seems that the enzymic procedure is un-satisfactory in that it does not effect the completehydrolysis of glycerides. Clement, Cl6ment-Champougny & Louedec (1954) found that themethod given by Kelsey (1939) did not effect morethan 67 % hydrolysis of the glycerides from various